AI for Performance Management in HR: A Review of Reported Effects on Evaluation and Development

This review synthesizes empirical evidence on AI techniques for performance management and employee development. We analyze reported outcomes from continuous monitoring systems, predictive modeling, and learning recommender engines. Documented effects include productivity increase (≈ 15%), on-time completion increase (≈ 12%), and engagement score increase (≈ 17%) after deployment of AI-supported feedback. Supervised classifiers for performance forecasting report accuracy (≈ 85%) with precision (≈ 83%) and recall (≈ 80%); AI metrics explain (≈ 60%) variance in satisfaction/retention. AI-guided internal mobility and development correlate with retention increase (≈ 12%). Methods include time-series analysis of key performance indicators, NLP for appraisal text and multi-source feedback, graph-based skill and role knowledge graphs, collaborative/contentbased recommenders, and retrieval-augmented LLM advisors. We summarize fairness auditing, inter-rater reliability gains, and explainability dashboards, and review privacy-preserving analytics via federated learning. The review consolidates evaluation pipelines, model validation practices, bias diagnostics, and governance mechanisms for decision support in HR, with evidence drawn solely from published implementations and observed results.